mirror of
https://github.com/searxng/searxng.git
synced 2024-12-29 12:40:30 +00:00
fd65c12921
languages.py can change, so users may query on a language that is not on the list anymore, even if it is still recognized by a few engines. also made no and nb the same because they seem to return the same, though most engines will only support one or the other.
127 lines
3.7 KiB
Python
127 lines
3.7 KiB
Python
"""
|
|
Swisscows (Web, Images)
|
|
|
|
@website https://swisscows.ch
|
|
@provide-api no
|
|
|
|
@using-api no
|
|
@results HTML (using search portal)
|
|
@stable no (HTML can change)
|
|
@parse url, title, content
|
|
"""
|
|
|
|
from json import loads
|
|
from urllib import urlencode, unquote
|
|
import re
|
|
from lxml.html import fromstring
|
|
|
|
# engine dependent config
|
|
categories = ['general', 'images']
|
|
paging = True
|
|
language_support = True
|
|
|
|
# search-url
|
|
base_url = 'https://swisscows.ch/'
|
|
search_string = '?{query}&page={page}'
|
|
|
|
supported_languages_url = base_url
|
|
|
|
# regex
|
|
regex_json = re.compile(r'initialData: {"Request":(.|\n)*},\s*environment')
|
|
regex_json_remove_start = re.compile(r'^initialData:\s*')
|
|
regex_json_remove_end = re.compile(r',\s*environment$')
|
|
regex_img_url_remove_start = re.compile(r'^https?://i\.swisscows\.ch/\?link=')
|
|
|
|
|
|
# do search-request
|
|
def request(query, params):
|
|
if params['language'] == 'all':
|
|
ui_language = 'browser'
|
|
region = 'browser'
|
|
elif params['language'].split('-')[0] == 'no':
|
|
region = 'nb-NO'
|
|
else:
|
|
region = params['language']
|
|
ui_language = params['language'].split('-')[0]
|
|
|
|
search_path = search_string.format(
|
|
query=urlencode({'query': query,
|
|
'uiLanguage': ui_language,
|
|
'region': region}),
|
|
page=params['pageno'])
|
|
|
|
# image search query is something like 'image?{query}&page={page}'
|
|
if params['category'] == 'images':
|
|
search_path = 'image' + search_path
|
|
|
|
params['url'] = base_url + search_path
|
|
|
|
return params
|
|
|
|
|
|
# get response from search-request
|
|
def response(resp):
|
|
results = []
|
|
|
|
json_regex = regex_json.search(resp.content)
|
|
|
|
# check if results are returned
|
|
if not json_regex:
|
|
return []
|
|
|
|
json_raw = regex_json_remove_end.sub('', regex_json_remove_start.sub('', json_regex.group()))
|
|
json = loads(json_raw)
|
|
|
|
# parse results
|
|
for result in json['Results'].get('items', []):
|
|
result_title = result['Title'].replace(u'\uE000', '').replace(u'\uE001', '')
|
|
|
|
# parse image results
|
|
if result.get('ContentType', '').startswith('image'):
|
|
img_url = unquote(regex_img_url_remove_start.sub('', result['Url']))
|
|
|
|
# append result
|
|
results.append({'url': result['SourceUrl'],
|
|
'title': result['Title'],
|
|
'content': '',
|
|
'img_src': img_url,
|
|
'template': 'images.html'})
|
|
|
|
# parse general results
|
|
else:
|
|
result_url = result['Url'].replace(u'\uE000', '').replace(u'\uE001', '')
|
|
result_content = result['Description'].replace(u'\uE000', '').replace(u'\uE001', '')
|
|
|
|
# append result
|
|
results.append({'url': result_url,
|
|
'title': result_title,
|
|
'content': result_content})
|
|
|
|
# parse images
|
|
for result in json.get('Images', []):
|
|
# decode image url
|
|
img_url = unquote(regex_img_url_remove_start.sub('', result['Url']))
|
|
|
|
# append result
|
|
results.append({'url': result['SourceUrl'],
|
|
'title': result['Title'],
|
|
'content': '',
|
|
'img_src': img_url,
|
|
'template': 'images.html'})
|
|
|
|
# return results
|
|
return results
|
|
|
|
|
|
# get supported languages from their site
|
|
def _fetch_supported_languages(resp):
|
|
supported_languages = []
|
|
dom = fromstring(resp.text)
|
|
options = dom.xpath('//div[@id="regions-popup"]//ul/li/a')
|
|
for option in options:
|
|
code = option.xpath('./@data-val')[0]
|
|
if code.startswith('nb-'):
|
|
code = code.replace('nb', 'no', 1)
|
|
supported_languages.append(code)
|
|
|
|
return supported_languages
|